Experimental Physics and
Industrial Control System

Andrew Johnson <[email protected]> · Mon, 17 Dec 2012 10:43:04 -0600

Hi Carlos,

On 2012-12-17 Carlos Pascual wrote:
> Are there any chances that something like this may end up being part of a
> base API in epics?

No.  As Mark Rivers has explained, most of the solutions people have been 
discussing just give you a list of record names.  Each record has something 
like a minimum of 40 different fields, so a typical 3.14 IOC with 1000 records 
loaded could have anything from 40,000 up to maybe 200,000 possible PV names 
that it would accept.  Now multiply that by the 500 or more IOCs that machines 
like the APS or SNS have and think about how much network bandwidth might be 
used up in just one query for "*" (each name can be up to 65 characters long).

Furthermore one of the directions that we are moving in for 3.15 is to add 
modifiable parameters to the PV name string, so for a 3.15 IOC the number of 
PV names that could point to just one record can't easily be calculated.

That's why large EPICS sites have built database systems like IRMIS or Ralph 
Lange's Channel Finder to make the current list of PVs available to clients or 
users without disturbing the real-time systems while making such queries.

In your case I would think the Channel Finder might be the most useful, but it 
does require that the names be loaded into the database.  There's no way 
around the fact that someone has to do the work somewhere.

HTH,

- Andrew
-- 
Computer science is as much about computers as astronomy is about
telescopes. -- Edsger Dijkstra

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System